I've been accumulating tube amplifier parts for many years and decided now that I'm retired I'd spend a little time and put something together. I like the lower distortion and better load dynamic load handling of push pull triode outputs so I was designing for that. I had a pair of 6000 to 4/8/16 ohm transformers that had been removed form a 30W Sansui integrated amplifier. I don't recall the model number, I bought them many years ago from a person that was parting it out. I had a number of 6BG6 sweep tubes including several that were very well matched that I was planning to use as my outputs but I could only get about 10 watts RMS from them and that wasn't quite enough for the speakers I have. My goal was 15 to 20 watts even though I realize that's not much more than 10, it is enough to listen to the speakers in my work area without them clipping.
I experimented with a couple other tubes but then decided to try the 6AR6 triode strapped. I'd read about them many years ago but had never actually designed anything with them. They are pretty readily available and are an inexpensive alternative to most other power tubes. I looked at their tube data curves and looked like I'd need on the order of 450 volts on the plate and a -75V bias to get 20 watts RMS in class AB1 with my transformers. So working backwards I needed about 150 volts peak to peak from the drive stage. I'm fortunate to have collected many power transformers over the years and I had one that was rated at 380-0-380 at 350 mA. That allowed me to use solid state rectifiers for the driver and get a 550V DC supply to regualte down to 525 and also get 450 from a tube rectifier with a CLC filter. I've had good results with the 6SL7 long tailed pair with a constant current bias and using 525 volts and large value (220K or larger) plate resisotrs. I follow that with a 6SN7 cathode follower so the bias resistors don't load the 6SL7. The 6SL7 is capable of a very linear output over a large (nearly 300V) output. One gain stage wasn't quite enough so I fed the long tailed pair with a 6SF5 triode. I could have used many different parts there but since the other tubes were octals I went with the 6SF5. It's not used that frequently but it has a mu of 100 and is much less expensive than a 12AX7 or 12AT7 and it is an octal base tube.
The inductor in the CLC filter is a 5H 300 mA choke, way larger then what I needed, but again I had it and it does have the benefit of low series resistance. I've tried both the 5V4 and 5U4 rectifiers on this amplifier, the 5V4 gives a higher DC voltage by about 15 volts under full load. At some point I'll try a 5AR4, but I only have one of those so I wanted to make sure eveything works first.
I did some breadboarding first and then when I was reasoanbly happy with the results I put this together. The driver stag ahs a voltage gain of approximately 65dB and the entire amplifier has an open loop THD under 5%. Closed loop, I can get 20 watts RMS per channel with 0.05% THD at 1 KHz and 0.15% at 20 KHz. At 5 watts out the THD is well under .05%. I got tired of setting up meters to adjsut the bias so I used inexpensive digital voltmeters across 10 ohm cathode resistors to measure and set the bias. I can adjust the level and balance for the outputs. I'm currently running them at 32 mA, I started at 40 mA but there was no advantage to that higher current.
I included a top and bottom view. The fuse was not yet connected when I took the picture.
I enjoyed building this amplifier and welcome and questions and comments. If anyone is interested I'll post the schematic.
I experimented with a couple other tubes but then decided to try the 6AR6 triode strapped. I'd read about them many years ago but had never actually designed anything with them. They are pretty readily available and are an inexpensive alternative to most other power tubes. I looked at their tube data curves and looked like I'd need on the order of 450 volts on the plate and a -75V bias to get 20 watts RMS in class AB1 with my transformers. So working backwards I needed about 150 volts peak to peak from the drive stage. I'm fortunate to have collected many power transformers over the years and I had one that was rated at 380-0-380 at 350 mA. That allowed me to use solid state rectifiers for the driver and get a 550V DC supply to regualte down to 525 and also get 450 from a tube rectifier with a CLC filter. I've had good results with the 6SL7 long tailed pair with a constant current bias and using 525 volts and large value (220K or larger) plate resisotrs. I follow that with a 6SN7 cathode follower so the bias resistors don't load the 6SL7. The 6SL7 is capable of a very linear output over a large (nearly 300V) output. One gain stage wasn't quite enough so I fed the long tailed pair with a 6SF5 triode. I could have used many different parts there but since the other tubes were octals I went with the 6SF5. It's not used that frequently but it has a mu of 100 and is much less expensive than a 12AX7 or 12AT7 and it is an octal base tube.
The inductor in the CLC filter is a 5H 300 mA choke, way larger then what I needed, but again I had it and it does have the benefit of low series resistance. I've tried both the 5V4 and 5U4 rectifiers on this amplifier, the 5V4 gives a higher DC voltage by about 15 volts under full load. At some point I'll try a 5AR4, but I only have one of those so I wanted to make sure eveything works first.
I did some breadboarding first and then when I was reasoanbly happy with the results I put this together. The driver stag ahs a voltage gain of approximately 65dB and the entire amplifier has an open loop THD under 5%. Closed loop, I can get 20 watts RMS per channel with 0.05% THD at 1 KHz and 0.15% at 20 KHz. At 5 watts out the THD is well under .05%. I got tired of setting up meters to adjsut the bias so I used inexpensive digital voltmeters across 10 ohm cathode resistors to measure and set the bias. I can adjust the level and balance for the outputs. I'm currently running them at 32 mA, I started at 40 mA but there was no advantage to that higher current.
I included a top and bottom view. The fuse was not yet connected when I took the picture.
I enjoyed building this amplifier and welcome and questions and comments. If anyone is interested I'll post the schematic.
Hello, the finished amp looks good!
I'm interested in the schematic for sure. 🙂
My concern is running 6AR6 in triode mode at 450V. I'm curious how well the tube's screen grid can take such a high voltage for a long period of time.
I'm interested in the schematic for sure. 🙂
My concern is running 6AR6 in triode mode at 450V. I'm curious how well the tube's screen grid can take such a high voltage for a long period of time.
I would be concerned with such a high voltage on the screen.
Do this . . .
Put a 100 Ohm resistor in series from the plate to the screen; very carefully use a handheld DMM to measure the voltage across the 100 Ohm resistor.
Name the 100 Ohm resistor: Rs
Volts Rs / 100 = the screen current, Is. (Amps, it will be fractions of an Amp)
Screen current, Is x 450V = Screen Dissipation.
The 6AR6 data sheet maximum screen dissipation of 3.2 Watts
3.2 Watts / 450V = 7.1 mA (0.0071 Amps).
0.0071A x 100 = 0.71VDC
You have to have less than 0.71VDC across the 100 Ohm resistor, or you have more than the maximum screen dissipation.
Please post a complete and accurate schematic. It will help me know if any of the ideas that I list below are good for the amplifier.
Solid state rectifiers increase the B+ voltage, versus tube rectifiers.
You have excess B+, do not use solid state rectifiers.
It appears there are 4 potentiometers (Bias adjust for each tube?)
Whatever the bias voltage is, the same voltage of Self Bias will subtract from the 450V B+ that is across the tube.
And self bias is less liable to have Thermal Run-Away of an output tube. In order to get the best current balance, and the best protection, you need individual self bias resistors.
Self Bias will reduce the output power a little. But you will not ever have to adjust the bias again.
Be sure to put a bypass capacitor across each of the self bias resistors.
The 5H 300mA choke is a good one for Choke input filter. But the B+ will drop too many volts.
So . . . just reduce the first capacitor capacitance of the CLC filter.
Perhaps it is 47uF, or 20uF for example. Replace that with a lower capacitance, try a 2uF or a 4uF cap, or perhaps more to get the B+ you want. The lower capacitance of the first cap will drop the B+ Volts.
Or, just wait and see what happens over time.
Just my thoughts of what I would do to the amplifier, for better reliability and longer life.
Do this . . .
Put a 100 Ohm resistor in series from the plate to the screen; very carefully use a handheld DMM to measure the voltage across the 100 Ohm resistor.
Name the 100 Ohm resistor: Rs
Volts Rs / 100 = the screen current, Is. (Amps, it will be fractions of an Amp)
Screen current, Is x 450V = Screen Dissipation.
The 6AR6 data sheet maximum screen dissipation of 3.2 Watts
3.2 Watts / 450V = 7.1 mA (0.0071 Amps).
0.0071A x 100 = 0.71VDC
You have to have less than 0.71VDC across the 100 Ohm resistor, or you have more than the maximum screen dissipation.
Please post a complete and accurate schematic. It will help me know if any of the ideas that I list below are good for the amplifier.
Solid state rectifiers increase the B+ voltage, versus tube rectifiers.
You have excess B+, do not use solid state rectifiers.
It appears there are 4 potentiometers (Bias adjust for each tube?)
Whatever the bias voltage is, the same voltage of Self Bias will subtract from the 450V B+ that is across the tube.
And self bias is less liable to have Thermal Run-Away of an output tube. In order to get the best current balance, and the best protection, you need individual self bias resistors.
Self Bias will reduce the output power a little. But you will not ever have to adjust the bias again.
Be sure to put a bypass capacitor across each of the self bias resistors.
The 5H 300mA choke is a good one for Choke input filter. But the B+ will drop too many volts.
So . . . just reduce the first capacitor capacitance of the CLC filter.
Perhaps it is 47uF, or 20uF for example. Replace that with a lower capacitance, try a 2uF or a 4uF cap, or perhaps more to get the B+ you want. The lower capacitance of the first cap will drop the B+ Volts.
Or, just wait and see what happens over time.
Just my thoughts of what I would do to the amplifier, for better reliability and longer life.
Last edited:
Here is one channel of the amplifier:
The filaments of U2 and U3 are offset 175 volts positive. That's shown on the power supply schematic. The 6SL7 operates at 1 mA per tube. The 60K resistors set the DC operation of the 6SN7 at 5 mA. Output tube bias is adjusted to 35 mA. I used 1K screen resistors and the DC current through them was about 2mA. Open loop gain on the driver stage only is about 65dB.
Here's the bias supply:
The bias adjust sets the level for both tubes and has a balance to adjust them to equal bias current. The balance control allows about a 5 volt difference between the DC set point on the tubes. Q1 is an emitter follower and R11 is a 2K potentiometer that adjusts the current in Q2 and Q3. The current changes the drop across the 2.7K resistors to set the balance. I know it seem complicated but it works well. The 90 volts is supplied by a separate bias transformer not shown on the schematic.
The HV power supply is shown below. The 380-0-380 volt transformer feeds a 5V4G tube rectifier and a solid state full wave rectifier. The tube rectifier has a CLC filter and a 450 volt output. The solid state feeds two 330uF caps in series and produces 550V which is regulated down to 525 for the driver. It alos has a tap and a pass FET for the filament offset.
The filaments of U2 and U3 are offset 175 volts positive. That's shown on the power supply schematic. The 6SL7 operates at 1 mA per tube. The 60K resistors set the DC operation of the 6SN7 at 5 mA. Output tube bias is adjusted to 35 mA. I used 1K screen resistors and the DC current through them was about 2mA. Open loop gain on the driver stage only is about 65dB.
Here's the bias supply:
The bias adjust sets the level for both tubes and has a balance to adjust them to equal bias current. The balance control allows about a 5 volt difference between the DC set point on the tubes. Q1 is an emitter follower and R11 is a 2K potentiometer that adjusts the current in Q2 and Q3. The current changes the drop across the 2.7K resistors to set the balance. I know it seem complicated but it works well. The 90 volts is supplied by a separate bias transformer not shown on the schematic.
The HV power supply is shown below. The 380-0-380 volt transformer feeds a 5V4G tube rectifier and a solid state full wave rectifier. The tube rectifier has a CLC filter and a 450 volt output. The solid state feeds two 330uF caps in series and produces 550V which is regulated down to 525 for the driver. It alos has a tap and a pass FET for the filament offset.
Thanks for posting the schematic.
The 2 mA Quiescent current on the screen does not cause the screen to be near the maximum dissipation specification.
When the signal drives the control grid more negative, and the plate voltage increases, the screen voltage increases along with the plate.
But since the control grid is more negative, the screen current does not increase. There is no excessive screen dissipation when the screen voltage swing is at maximum.
If the screen never arcs when the plate and screen voltage are at maximum swing, the output tubes are safe.
It seems that is not a problem, you have already taken the amplifier through its tests and listening conditions.
Looks good to me.
Your application of the 6AR6 reminded of the 807 beam power tube's original maximum screen voltage specification, 300V.
Later, designers started using the 807 in Triode mode, a new specification was added: Triode Mode, maximum quiescent screen voltage: 400V.
One Early Williamson amplifier designs may have been one of the first to use the 807 in Triode mode.
The 2 mA Quiescent current on the screen does not cause the screen to be near the maximum dissipation specification.
When the signal drives the control grid more negative, and the plate voltage increases, the screen voltage increases along with the plate.
But since the control grid is more negative, the screen current does not increase. There is no excessive screen dissipation when the screen voltage swing is at maximum.
If the screen never arcs when the plate and screen voltage are at maximum swing, the output tubes are safe.
It seems that is not a problem, you have already taken the amplifier through its tests and listening conditions.
Looks good to me.
Your application of the 6AR6 reminded of the 807 beam power tube's original maximum screen voltage specification, 300V.
Later, designers started using the 807 in Triode mode, a new specification was added: Triode Mode, maximum quiescent screen voltage: 400V.
One Early Williamson amplifier designs may have been one of the first to use the 807 in Triode mode.
Last edited: